PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Am J Transplant. Author manuscript; available in PMC 2011 December 1.
Published in final edited form as:
PMCID: PMC3059496
NIHMSID: NIHMS246079

Height Contributes to the Gender Difference in Wait-list Mortality Under the MELD-Based Liver Allocation System

Jennifer C. Lai, MD, MBA,a Norah A. Terrault, MD, MPH,a Eric Vittinghoff, PhD, MPH,b and Scott W. Biggins, MD, MASc

Abstract

This study examined factors associated with the gender disparity in wait-list mortality in the MELD era. Adult patients listed for liver transplantation from 2002-2008 were included. Females [12,585(36%)] and males [22,126(64%)] differed clinically by age (54 vs. 52 years), height (1.6 vs. 1.8 meters), listing estimated glomerular filtration rate [eGFR; 70 vs. 83 ml/min], and cirrhosis etiology. Holding MELD constant, females were at 19% (95%CI, 1.13-1.25, p<0.001) higher risk of wait-list mortality than males under the current allocation system. The relative hazard increased with worsening renal function, whether measured by serum creatinine or eGFR. Adjustment for MELD, age, African-American race, cirrhosis etiology, region, and ABO group attenuated this relative hazard (HR 1.16; 95%CI, 1.10-1.22; p<0.001) but additional adjustment for height completely explained this gender disparity in wait-list mortality (HR 1.05; 95%CI, 0.98-1.12; p=0.2). Transplantation rates, however, remained lower among females, even after adjustment for height (HR 0.88; 95%CI, 0.82-0.92; p<0.001). In conclusion, under the current liver allocation system, women have a 19% increased risk of wait-list mortality compared to men with the same MELD scores. Height contributes to this gender disparity, possibly reflecting differences in transplantation rates for shorter individuals.

BACKGROUND

In an effort to prioritize patients with end-stage liver disease with the highest short-term mortality, the United Network for Organ Sharing (UNOS) implemented a liver allocation system based on the Model for End-Stage Liver Disease (MELD) score in 2002.(1) The MELD score, derived from three objective routine laboratory tests – total bilirubin, international normalized ratio for prothrombin time (INR), and creatinine(2) – has been validated as an accurate predictor of mortality in patients awaiting liver transplantation.(3-5) Prior to this, donor livers were allocated based on a combination of laboratory values, clinical assessment, and wait-list time, a system that was criticized for its relative subjectivity. Since its implementation, the MELD score liver allocation system has been shown to effectively decrease median wait-list times and wait-list mortality for liver transplant candidates.(6)

Despite these improvements, the MELD score appears to be associated with a new disparity – one that is based on gender. A recent study comparing all patients listed for liver transplantation in a pre-MELD versus a post-MELD cohort found that females were 9% more likely to die or become too sick for transplant on the wait-list compared to males in the post-MELD era, whereas there was no gender difference in the pre-MELD era.(7)

The cause of this gender difference remains unclear. Several experts have suggested that this disparity in wait-list mortality reflects differences in body size and renal function, but no study has demonstrated the association between gender-related factors and outcomes on the waiting list.(8, 9) Therefore, we designed this study specifically to evaluate whether factors that differ by gender are associated with the increased liver transplant wait-list mortality in females compared with males.

METHODS

Patients

All patients who were listed for single-organ liver transplantation in the United States from March 1, 2002 through December 31, 2008 with complete data were evaluated for inclusion in this study. Patients who were less than 18 years old, listed for retransplantation or as Status 1, including those with fulminant hepatic failure or acute hepatic necrosis, were excluded. As serum creatinine fluctuates with renal replacement therapy and therefore, may not accurately reflect renal function (i.e., lower values cannot be interpreted as improving renal function), patients who had received dialysis were also excluded from our analyses. Given their differing likelihood of receiving a transplant, patients who were listed with exception points for any reason, including but not limited to hepatocellular carcinoma (HCC), and those who underwent living donor liver transplantation were excluded.

Covariates

All data were collected from the United Network for Organ Sharing (UNOS) registry. Data included gender, age at listing, race, height, weight, ABO group, UNOS region (1 through 11), date of listing, listing diagnoses, total bilirubin at listing, INR at listing, serum creatinine at listing, receipt of dialysis, death date, date of removal from the waiting list, reason for removal, transplant date, and type of transplant. Cutoffs that were deemed to be implausible were as follows: height <120 cm and >240 cm, weight <30 kg and >180 kg, total bilirubin ≤0 mg/dL, INR ≤0, creatinine ≤0 mg/dL. Body mass index (BMI) was calculated using the formula: weight (kilograms) / height (meters)2. Estimated glomerular filtration rates (eGFR) were calculated using the Modification for Diet in Renal Disease (MDRD) equation.(10) The MELD score was calculated using the standard formula(3):

MELD = 3.8*ln(bilirubin[mg/dL]) + 11.2*ln(INR) + 9.6*ln (creatinine[mg/dL]). For the calculation of MELD at transplant, a lower limit of 1 was set for all variables. In order to capture the full association between MELD score and mortality in this cohort, no upper limit was set for the MELD score.

Although patients of all races were included in this study, race, as a covariate in our analysis, was grouped as African-American versus non-African-American. UNOS region of listing was categorized by mean MELD score at transplantation into low (regions 3,6,10,11; mean MELD range 22.1-22.9), medium (regions 2,4,7,8; mean MELD range 23.3-25.5), and high (regions 1,5,9; mean MELD range 26.2-28.2) risk regions. Listing diagnoses were grouped into the following common diagnostic categories: hepatitis C virus (HCV), hepatitis B virus (HBV), non-alcoholic fatty liver disease (NAFLD, including cryptogenic cirrhosis and non-alcoholic steatohepatitis), alcoholic cirrhosis, autoimmune etiologies (including primary biliary cirrhosis, primary sclerosing cholangitis, and autoimmune hepatitis), and other etiologies of cirrhosis (including alpha-1 antitrypsin deficiency and hemochromatosis). Patients who were listed with HCV in addition to other diagnoses were included under a listing diagnosis of HCV.

Outcomes and Censoring

The focus of interest was the association between gender and the risk of death or becoming too sick for transplant. Patients who reached this combined endpoint (“wait-list mortality”) were those whose reason for removal was “died”, “medically unsuitable”, or “too sick for transplant”. Patient follow-up began on the date of listing and ended at the time of death, removal from the waiting list, transplant, or date of last data update on the UNOS registry. When patients had multiple listings, the earliest listing event was used.

Statistical Analysis

The characteristics of females and males at listing were compared using Wilcoxon and chi-square tests as appropriate. Competing risk analysis, which has been described in detail by Kim et al(11), was used in all analyses assessing the association between gender and the combined endpoint of death or becoming too sick for transplant on the waiting list. In this analysis using the Fine-Gray model, transplantation and/or removal from the waiting list for other reasons were treated as competing risks, because death on the waiting list cannot logically occur after either of these events.(12) Patients who remained alive and on the waiting list at the end of follow-up were conventionally censored at the date of the most recent update of the UNOS registry, since they remained at risk for wait-list mortality. In contrast to the Cox model, this analysis is sensitive to indirect adverse or protective effects: for example, a factor associated with the competing risk of early transplantation could appear protective against death on the waiting list, by reducing the time at risk. Similarly, factors associated with delayed transplantation – in our analysis, height may play this role – may indirectly put patients at higher risk of wait-list mortality. Adjusted cumulative incidence was also estimated based on the competing risk model. This competing risks approach was also used to assess the influence of gender and height on transplantation rates.

To determine which covariates to include in the multivariate models, we used unadjusted models to assess the effects of MELD, age, height, African-American race, etiology of cirrhosis, body mass index, region risk category, and ABO group, and also exhaustively screened for interactions between gender, age, African-American race, serum creatinine, and height. Covariates and interactions with p-values ≤ 0.05 were included in multivariate models. Multi-collinearity was checked using a variance inflation factor cutoff of ≥10.(13) The non-linear relationship between MELD scores and wait-list mortality was modeled using a restricted cubic spline.(14) In the models evaluating the role of renal function, MELD was modeled using two cubic splines, one for the sum 3.8*ln(bilirubin) + 11.2*ln(INR), and the other for (9.6*ln(creatinine)) or ln(eGFR).

Two-sided p-values ≤ 0.05 were considered statistically significant. Analyses were performed using Stata®11.0 statistical software (College Station, Texas). This study was approved by the institutional review board at the University of California San Francisco.

RESULTS

From March 1, 2002 through December 31, 2008, 61,749 patients were listed for single-organ liver transplantation through the UNOS registry. Of these patients, 12,585 (36%) females and 22,126 (64%) males met the inclusion criteria (Figure 1). Characteristics of the study cohort of 34,711 liver transplant candidates are shown in Table 1. At listing, females were older, of shorter height and lower weight, and more likely to be of African-American or Hispanic race. A significantly greater proportion of females were transplanted in UNOS regions with medium or high mean MELD scores at transplantation. Females were more likely to be listed for NAFLD or autoimmune etiologies of liver disease. Serum creatinine, eGFR, and MELD scores at listing also differed.

Figure 1
Flow of Patients Listed for Liver Transplantation. Abbreviations: HCC, hepatocellular carcinoma; FHF, fulminant hepatic failure; AHN, acute hepatic necrosis; Cr, serum creatinine at listing.
Table 1
Characteristics of Patients Listed for Liver Transplantation in the United States from March 1, 2002 through December 31, 2008

During the follow-up period, 2,770 (22%) females and 4,340 (20%) males died or became too sick for transplantation; 4,894 (39%) females and 10,059 (46%) males received a deceased donor liver transplant; and 1,359 (11%) females and 2,241 (10%) males were removed for other reasons.

Cumulative incidence of wait-list mortality by gender is shown in Figure 2. In this initial analysis adjusting only for MELD score, females were at an estimated 19% (HR, 1.19; 95% CI, 1.13-1.25; p<0.001; Figure 3) increased risk of dying or becoming too sick for transplant, compared with males. Adjustment for MELD, age, African-American race, etiology of cirrhosis, region risk group, ABO group, and significant interactions between these covariates explained only a small part of the association between gender and wait-list mortality risk (Figure 3). However, additional adjustment for height almost completely explained the remaining association (Figure 3).

Figure 2
Cumulative incidence curves comparing the risk of wait -list mortality in females compared with males, adjusted for MELD only.
Figure 3
Adjusted Risk of Wait-list Mortality for Females Compared with Males.

The majority of women were distributed in the lowest quartile of the height distribution of wait-list candidates (Figure 5). Compared to patients in the uppermost quartile of height, mortality rates were 8% higher in the middle two quartiles (HR, 1.08; 95% CI, 1.02-1.15; p=0.01), and 24% higher in the lowest quartile of height (HR, 1.24; 95% CI, 1.14-1.36; p<0.001).

Figure 5
Height of Female and Male Wait-List Candidates in the United States.

In an exploratory analysis, adjustment for weight rather than height as a marker for body size explained only a small part of the adjusted association with gender (adjusted HR, 1.13; 95% CI, 1.07-1.19; p<0.001). Further exploratory analysis revealed that the relative hazard of wait-list mortality in females compared to males was larger in patients listed with worse renal function, whether estimated by serum creatinine or eGFR. Specifically, after adjusting for INR and bilirubin, the relative hazard of wait-list mortality by gender increased with increases in creatinine (p for interaction <0.001; Figure 4A). Similarly, decreases in eGFR, representing worse renal function, were associated with increases in the gender disparity in wait-list mortality (Figure 4B; p for interaction <0.001).

Figure 4
Relative Hazard of Wait-List Mortality for Females Compared to Males By A) Creatinine at Listing (mg/dL) and B) GFR at Listing (ml/min).

We next explored whether delays in transplantation specific to height or gender could explain our findings. In a competing risks analysis of transplantation, rates were 17% lower among women (HR, 0.83; 95% CI, 0.80-0.86; p<0.001) after adjustment for MELD, age, region, and blood group. After further adjustment for quartile of height, adjusted transplantation rates remained 12% lower among women (HR, 0.88; 95% CI, 0.83-0.92; p<0.001). As compared to patients in the uppermost quartile of height, transplantation rates were 3% lower (HR, 0.97; 95% CI, 0.93-1.01; p=0.12) in the middle two quartiles and 11% lower (HR, 0.89; 95% CI, 0.83-0.95; p<0.001) in the shortest quartile.

To further investigate the association between gender and height on wait-list events, median MELD scores at each event were evaluated (Table 2). Female wait-list candidates in the short and average height categories were listed with a lower median MELD score compared to male wait-list candidates, but died at similar median MELD scores. In contrast, short and average height female candidates were transplanted at a slightly lower MELD score than males of similar heights, although this only met the pre-specified threshold for statistical significance (p<0.05) within the average height category. There was a statistically significant gender difference in the median MELD score at the time of other removal among those of average height.

Table 2
MELD Scores by Height Categories at Listing, Death, Transplant, and Other Removal

DISCUSSION

The clinical benefits of the objective MELD-based liver allocation system over the prior system that incorporated waiting time with subjective clinical assessments of illness are clear. Within 4 years of its adoption, median waiting time declined by nearly 70% and death rates declined by 20%.(15) Recently, Moylan and colleagues(7) reported that this new system had eliminated race-based disparities documented in the pre-MELD era.(16) However, in their UNOS-based cohort of African-American and white patients only, Moylan et al found a new gender disparity. Our competing risks analysis in a MELD-era cohort including all races also found increased wait-list mortality risk among women.

Importantly, we found that women were listed with a lower MELD score despite worse renal function as calculated using eGFR relative to males. Cholangitas et al(17) and Huo et al(18) have both previously demonstrated that using a creatinine value that is “corrected” for eGFR increases the MELD score for the majority of women by two or three points. This suggests that gender disparities in liver allocation could result from insensitivity of the creatinine-based MELD score to gender differences in renal function. However, adjusting for eGFR, which accounts for the gender difference in renal function, rather than creatinine, did not substantially explain the higher risk of wait-list mortality among women. Furthermore, the elevated risk is present among most women, excepting only those with low creatinine or high eGFR. Our analyses underscore the importance of the development and implementation of more accurate markers of renal function, such as cystatin C, into our national liver allocation system.

While adjustment for age, African-American race, etiology of cirrhosis, UNOS region risk, and ABO group explained only a small fraction of the gender disparity, further adjustment for height explained most of the rest. This is an important finding as women were, on average 15 cm shorter than men, suggesting that the increased mortality in shorter patients is the main driver of this gender disparity in wait-list mortality.

In a sensitivity analysis, weight did considerably less than height to explain the gender disparity in mortality rates. Weight can vary dramatically due to complications of end-stage liver disease, with one patient gaining weight because of ascites and another patient losing secondary to cachexia, making weight a “noisy” surrogate for acceptable donor liver size.

The association between height and wait-list mortality is not clearly understood. We explored several hypotheses. One potential explanation is that shorter women need smaller organs, which are preferentially offered to pediatric recipients. As a result, shorter women at the top of the waiting list may have to wait longer for a size-appropriate liver graft. However, our analysis only partly supports this explanation. Although shorter height explains most of the increased wait-list mortality among women and is associated with both higher mortality and lower transplantation rates, adjusted transplantation rates remained lower among women even after accounting for height. This suggests that factors in addition to lower availability of size-appropriate grafts contribute to the lower transplantation rates and higher wait-list mortality among women.

We also considered whether differences by height categories at median MELD scores at the time of listing, death, or transplantation might explain the association between height and wait-list mortality. Women in the short and average height categories were listed with a lower MELD score but died at a similar MELD score compared to men. However, there are several factors other than MELD score that influence the likelihood of dying on the waiting list such as listing region and blood group. The multivariate competing risks analysis, which accounts for these factors, in addition to accounting for the competing risks of transplantation or removal for other reasons, is therefore the most accurate method of assessing predictors of wait-list mortality. We show that height remains an important predictor of wait-list mortality, despite accounting for all of these factors.

Lastly, we considered whether the association between wait-list mortality and height was related to an interaction between renal function and height, providing an explanation for the findings in our study. We found no evidence that the association between serum creatinine and wait-list mortality varies by height. We hypothesize that, in patients with end-stage liver disease, who generally have very low muscle mass regardless of height, serum creatinine values do not differ as greatly between patients of different heights (with the same renal function) as they would in healthy patients. Therefore, further studies are needed to explore other factors that might explain the association between height, gender, and wait-list mortality such as differential causes of death or rates of disease progression.

Why might this gender disparity have become apparent in the post-MELD, but not the pre-MELD, era? We offer two reasons. First, women may have been listed at an earlier stage in their disease process by the relatively subjective Child classification system than they are now using the more objective MELD-based system. Therefore, women no longer benefit from early listing in the MELD-era. Second, the inclusion of serum creatinine in the liver allocation system may have allowed this gender disparity in wait-list mortality to now become apparent.

In conclusion, our study is the first to show that shorter individuals have a higher wait-list mortality, and women are disproportionately affected. The strong association that we found between body size and the gender disparity in wait-list mortality represents an important area for future refinements to the national liver allocation system. Given the severe deficit of donor organs relative to their need, continued investigation within these areas is essential to ensuring the most effective allocation of this scarce resource.

Acknowledgements

We would like to thank Dr. David V. Glidden, Professor of Biostatistics at the University of California, San Francisco for his expert input on survival analysis.

Grant support: This project was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (T32 DK060414, JCL; DK076565, SWB), the National Center for Research Resources (KL2 RR024130, SWB), the Agency for Healthcare Research and Quality (DK076565, SWB), and the University of California San Francisco Liver Center (JCL, NAT, SWB).

Footnotes

Disclosures: Nothing to disclose.

REFERENCES

1. Freeman RB, Jr., Wiesner RH, Harper A, McDiarmid SV, Lake J, Edwards E, et al. The new liver allocation system: moving toward evidence-based transplantation policy. Liver Transpl. 2002 Sep;8(9):851–8. [PubMed]
2. Malinchoc M, Kamath PS, Gordon FD, Peine CJ, Rank J, ter Borg PC. A model to predict poor survival in patients undergoing transjugular intrahepatic portosystemic shunts. Hepatology. 2000 Apr;31(4):864–71. [PubMed]
3. Kamath PS, Wiesner RH, Malinchoc M, Kremers W, Therneau TM, Kosberg CL, et al. A model to predict survival in patients with end-stage liver disease. Hepatology. 2001 Feb;33(2):464–70. [PubMed]
4. Wiesner RH, McDiarmid SV, Kamath PS, Edwards EB, Malinchoc M, Kremers WK, et al. MELD and PELD: application of survival models to liver allocation. Liver Transpl. 2001 Jul;7(7):567–80. [PubMed]
5. Wiesner R, Edwards E, Freeman R, Harper A, Kim R, Kamath P, et al. Model for end-stage liver disease (MELD) and allocation of donor livers. Gastroenterology. 2003 Jan;124(1):91–6. [PubMed]
6. Freeman RB, Wiesner RH, Edwards E, Harper A, Merion R, Wolfe R. Results of the first year of the new liver allocation plan. Liver Transpl. 2004 Jan;10(1):7–15. [PubMed]
7. Moylan CA, Brady CW, Johnson JL, Smith AD, Tuttle-Newhall JE, Muir AJ. Disparities in liver transplantation before and after introduction of the MELD score. JAMA. 2008 Nov 26;300(20):2371–8. [PMC free article] [PubMed]
8. Dureja P, Lucey MR. Disparities in liver transplantation in the post-model for end-stage liver disease era: are we there yet? Hepatology. 2009 Sep;50(3):981–4. [PubMed]
9. Axelrod DA, Pomfret EA. Race and sex disparities in liver transplantation: progress toward achieving equal access? JAMA. 2008 Nov 26;300(20):2425–6. [PubMed]
10. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med. 1999 Mar 16;130(6):461–70. [PubMed]
11. Kim WR, Therneau TM, Benson JT, Kremers WK, Rosen CB, Gores GJ, et al. Deaths on the liver transplant waiting list: an analysis of competing risks. Hepatology. 2006 Feb;43(2):345–51. [PubMed]
12. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Amer Stat Assoc. 1999;94(446):496–509.
13. Chatterjee S, Hadi AS. Regression analysis by example. 4th edition Wiley; New York: 2006.
14. Harrell FE. Regression modeling strategies. Springer; New York: 2001.
15. Freeman RB, Jr., Steffick DE, Guidinger MK, Farmer DG, Berg CL, Merion RM. Liver and intestine transplantation in the United States, 1997-2006. Am J Transplant. 2008 Apr;8(4 Pt 2):958–76. [PubMed]
16. Reid AE, Resnick M, Chang Y, Buerstatte N, Weissman JS. Disparity in use of orthotopic liver transplantation among blacks and whites. Liver Transpl. 2004 Jul;10(7):834–41. [PubMed]
17. Cholongitas E, Marelli L, Kerry A, Goodier DW, Nair D, Thomas M, et al. Female liver transplant recipients with the same GFR as male recipients have lower MELD scores--a systematic bias. Am J Transplant. 2007 Mar;7(3):685–92. [PubMed]
18. Huo SC, Huo TI, Lin HC, Chi CW, Lee PC, Tseng FW, et al. Is the corrected-creatinine model for end-stage liver disease a feasible strategy to adjust gender difference in organ allocation for liver transplantation? Transplantation. 2007 Dec 15;84(11):1406–12. [PubMed]